Why I Worry About Computerized Voting

Marian Beddill is a retired civil engineer who worked on water-related projects for most of her professional career, and tied that work into computers and mapping since the mid-1960s. In Bellingham since 1991, she has been active in many civic issues, especially the environment (http://LakeWhatcom.org) and good government (http://NoLeakyBuckets.org).

People who write (and certify) programs for computers can make mistakes. Such mistakes can make the results come out wrong. Wrong results can change the outcome of elections. That’s about as simple as I can make it.

Some mistakes end up being ho-hum kinds of things, and the situation can be fixed with either a temporary workaround or a permanent correction. But others may be worse, might not be easily patched up, and may make a difference in the election outcome — an intolerable result. Worse yet, they might make changes in the outcome that do not get caught — or not caught until “too late.” 

Here is a current, real case, of such a computer program mistake, in my own county (Whatcom).

We use vote-by-mail. Ballots are printed and mailed to voters. Voters mark the ballot and mail them back (or drop the ballot in a designated collection box.) Whatcom County is geographically divided two different ways — by precincts and by special taxing-districts, like school, fire, cemetery and parks districts. Their boundaries don’t coincide; many precincts have sub-sections commonly called “splits.” Voters in one split can vote on some offices and measures; voters in other splits cannot.

Our new (three years old) computerized elections system has multiple programs for the many different functions. And those programs must interact with each other, and must do so flawlessly. If they do not correctly interact, blam! Even more troubling, they (and the fundamental “operating system” of the computer) are made by different companies, so the coordination is tricky or worse. 

This year, one of the companies changed their program. The company who made it either didn’t know about, or didn’t adequately check, how it interacted with another program. The result was a failure. And yet the changed program (we are told) was passed — approved and certified by the federal certification process and the state certification process. Why was the error not caught there?

So, as our county was preparing the ballot-images for our OpScan ballots to be sent to the printer, two of the programs goofed in interacting, and the computer output had errors. The program that was changed had a new identification way of registering the precinct-split number, but the other program that received that data did not know the new field identification. 

The outcome of that goofed merger? The information was ignored, and the ballots were printed without the split identification — 9,211 of them (almost 10 percent of the county’s approximately 100,000 registered voters) and mailed to the voters, sans the computerized bar-coded split IDs.

Only when the next step was being done, setting up the ballot-reading program, did a third program squawk — because it saw that those IDs were missing. Aw, shucks. It took the staff several days to trace the problem, and to craft a solution.

Maybe that will get fixed well enough for this election, already running. And surely well enough for the next election — until the next mistake appears.

This is the main reason I do not have a high degree of confidence in computerized voting and vote counting. The potential — nay, certainty — of errors is too high to engender the degree of confidence needed, thus the trust that the public must have, in the use of elections as the way to let the people control the government.

The opposite is intolerable, and the founders of our nation knew that, and wrote it into the Constitution of the United States.

Maybe we can continue to use computers in our elections systems, but only with a parallel auditing program that can be independently verified totally separate from the basic system.

And since I reject the use of yet other computers to do this, the only such process which I know to check the accuracy and integrity of the vote counting is by hand and eye. This is why I strongly support the standard use, everywhere and always, of a program of hand-counted double-checking of the ballots and vote counting. 

Because: “If you cannot trust the way your votes are counted, nothing much else in politics matters!” And we do want politics to matter; the alternative is not democracy. §